Multi-hop whistler-mode ELF//VLF signals and triggered emissions

excited by the HAARP HF heater

U. S. Inan,1 M. Gol-kowski,1 D. L. Carpenter,1 N. Reddell,1,2 R. C. Moore,1 T. F. Bell,1

E. Paschal,3 P. Kossey,4 E. Kennedy,5 and S. Z. Meth6

Received 30 September 2004; revised 13 November 2004; accepted 24 November 2004; published 28 December 2004.

[1] Modulated heating of the lower ionosphere with theHAARP HF heater is used to excite 1–2 kHz signalsobserved on a ship-borne receiver in the geomagneticconjugate hemisphere after propagating as ducted whistler-mode signals. These 1-hop signals are believed to beamplified, and are accompanied by triggered emissions.Simultaneous observations near (�30 km) HAARP show2-hop signals which travel to the northern hemisphere uponreflection from the ionosphere in the south.Multiple reflectedsignals, up to 10-hop, are detected, with the signal dispersingand evolving in shape, indicative of re-amplification and re-triggering of emissions during successive traversals ofthe equatorial interaction regions. INDEX TERMS: 2403

Ionosphere: Active experiments; 2483 Ionosphere: Wave/particle

interactions; 2736 Magnetospheric Physics: Magnetosphere/

ionosphere interactions; 2794Magnetospheric Physics: Instruments

and techniques. Citation: Inan, U. S., M. Gol-kowski, D. L.

Carpenter, N. Reddell, R. C. Moore, T. F. Bell, E. Paschal,

P. Kossey, E. Kennedy, and S. Z. Meth (2004), Multi-hop

whistler-mode ELF/VLF signals and triggered emissions excited

by the HAARP HF heater, Geophys. Res. Lett., 31, L24805,

doi:10.1029/2004GL021647.

1. Introduction

[2] Electromagnetic waves in the 4 Hz to 6.5 kHz rangeare known to be generated by modulated HF heating of thelower ionosphere through which auroral electrojet currentsflow [Barr and Stubbe, 1984; Villasenor et al., 1996]. ELF/VLF waves have been generated at the High FrequencyActive Auroral Research Program (HAARP) in Gakona,Alaska using HF heating modulated at ELF/VLF, under awide range of geomagnetic conditions (R. C. Moore et al.,ELF/VLF waves generated by an artificially modulatedauroral electrojet above the HAARP HF heater, submittedto Journal of Geophysical Research, 2004, hereinafterreferred to as Moore et al., submitted manuscript, 2004).HAARP is located at L ’ 4.9, where the magnetic fieldlines are usually dipole-like and tend to lie within or nearthe plasmaspause. HAARP is thus well positioned for use in

controlled wave-injection experiments to study ELF/VLFwave growth and emission triggering, similar to those con-ducted during 1973–88 with the Siple Station, AntarcticaVLF transmitter (L ’ 4.2). Siple Station [Helliwell, 1988]consisted of a�100 kW transmitter (later �150 kW) drivinga 21 km horizontal antenna (later extended to 42 km crosseddipoles) placed on an ice sheet �2 km in thickness. Thetransmitter launched �1.6 to 5 kHz waves on field linesranging from L = 3 to L = 5 observed at receivers in thegeomagnetically conjugate region in Canada, with ducting,amplification, and emission triggering occurring in manycases. We report here the first observations of the excitationby an HF heater of ducted whistler-mode ELF/VLFsignals, amplified in the magnetosphere and accompaniedby triggered emissions (Figure 1).

2. Review of HF Heater ELF/VLF Generation

[3] The EISCAT HF heater near Tromsø, Norway hasbeen used to generate ELF/VLF signals [Stubbe et al., 1982;Barr and Stubbe, 1984, 1991; Rietveld et al., 1989] withamplitudes of �1 pT on the ground. With a total radiatedHF power of 1 MW and effective radiated power (ERP) of200–300 MW at 2.75 to 8 MHz, the Tromsø heater wasoften 100% amplitude modulated with a square wave.Tromsø is located at L > 6, and is thus on sub-auroral/auroral field lines on which conditions for hemisphere-to-hemisphere ducting are less favorable [Carpenter and Sulic,1988]. HF ionospheric heaters at Arecibo, HIPAS, andHAARP have been used to modulate ionospheric currentsystems. At Arecibo, 500 Hz to 5 kHz waves were producedusing �3 MHz with a total HF input power of 800 kW, andan ERP of 160–320 MW [Ferraro et al., 1982]. The HFheater may have sometimes created field aligned ducts[Starks et al, 2001]. At HIPAS, ELF/VLF waves werecreated using amplitude and phase modulation, mostsuccessfully when the electrojet was in the path of the HFbeam, when there was low D region absorption, and whenenergetic particle precipitation and visible aurora were notoverhead [Villasenor et al., 1996]. ELF/VLF wave genera-tion at HAARP was found to be most efficient [Milikh et al.,1999] for �3.3 MHz in X-mode with 100% square wavemodulation.

3. Experimental Setup

3.1. The High Frequency Active AuroralResearch Program (HAARP)

[4] The HAARP HF heater is located at �62.4�N and145.2�W geographic (63.1�N and 92.4�W geomagnetic). Ahigh power, HF phased-array transmitter is used to heatsmall, well-defined volumes of the ionosphere (see:

GEOPHYSICAL RESEARCH LETTERS, VOL. 31, L24805, doi:10.1029/2004GL021647, 2004

1Space, Telecommunications, and Radioscience (STAR) Laboratory,Stanford University, Stanford, California, USA.

2Now with United States Navy, Saratoga Springs, New York, USA.3Whistler Radio Services, Anderson Island, Washington, USA.4Air Force Research Laboratory, Hanscom Air Force Base, Massachu-

setts, USA.5Naval Research Laboratory, Washington, D. C., USA.6Defense Advanced Research Programs Agency (DARPA), Arlington,

Virginia, USA.

Copyright 2004 by the American Geophysical Union.0094-8276/04/2004GL021647$05.00

L24805 1 of 4

www.haarp.alaska.edu). The transmitter had a total radiatedHF power capability of 960 kW at the time of the observa-tions. HF carriers of 3.25 MHz and 5.8 MHz were used, withthe ELF/VLF signal format impressed as 100% sinusoidalamplitude modulation.

3.2. ELF/VLF Receivers

[5] The ELF/VLF receivers utilized large square (4.8 mby 4.8 m) or triangular shipboard (4.2 m high with 8.4 mbase) air core antennas, with terminal resistive and inductiveimpedances respectively of 1-W and 1-mH, matched toa low-noise (noise figure of �2–3 dB at a few kHz)preamplifier with a flat frequency response (�300 Hzto �40 kHz). The data in the north were acquired atChistochina, Alaska, within �35 km of HAARP. Observa-tions in the south were conducted on the research vesselTangaroa, while it was near (within �100 km) the geo-magnetically conjugate point of HAARP, to deploy a buoyfor autonomous ELF/VLF measurements of 1-hop ductedwhistler-mode signals excited by HAARP.

4. Observations

[6] Between 0200UT and 1500UTeach day from April 19to April 26, 2004, HAARP transmitted continuously,repeating a 1-minute long ELF/VLF modulation consistingof a sequence of frequency-time ramps, pulses, and chirps.HF transmissions were in the X-mode, alternating between3.25 MHz and 5.8 MHz every 30-minutes, operating at fullpower (960 kW), with the HF beam oriented vertically.[7] The multi-hop ducted whistler-mode signals were

observed on 20 April 2004 between 0310 and 0345 UT.No evidence for whistler-mode echoes was observed ineither hemisphere outside of this �1 hour period. Figure 2shows two well defined examples of 1-hop signals observedon the Tangaroa, located at 55.38�S and 174.65�E (L’ 4.5).The top panel shows relatively weak HAARP transmittedfrequency-time ramps and pulses observed at Chistochina inaddition to natural activity in the range �1 to �2 kHz. Thesecond and third panels show Tangaroa data during the first15 seconds of two successive minutes. The 1-hop signal isvisible at the same time on both panels. The repeatedoccurrence of this signal at the same time (in the twominutes shown and in others not shown) is clear evidenceof a causal connection to the HAARP transmissions. The

steepening of the frequency-time slope of the 1-hop signalwith respect to that of the transmitted ramp (visible in thetop panel) is generally consistent with whistler-mode dis-persion (in this case leading to signal compression) from�1 to �2 kHz. These frequencies are below the so-called‘nose’ frequency of fastest travel [Helliwell, 1965, p. 32] atthe inferred L-shell of �4.9. Figure 2 (top panel) also showsthe 2-hop echo, delayed (at each frequency) from the 1-hop(on the Tangaroa records) by as much as the 1-hop signal is

Figure 1. Schematic of ducted whistler-mode propagationexcited by the HAARP HF heater.

Figure 2. Spectrograms showing 2-hop and 1-hop echoesreceived at Chistochina, Alaska (top panel) and at themagnetic conjugate point on the RV Tangaroa (two bottompanels).

Figure 3. Spectrograms showing HAARP transmissionformat (top panel), echoes recorded at Chistochina (middlepanel) and echoes recorded on the RV Tangaroa (bottompanel).

L24805 INAN ET AL.: EXCITATION OF DUCTED ELF/VLF SIGNALS BY HAARP L24805

2 of 4

delayed with respect to the parent frequency-time ramp. Theincreased steepness of the 2-hop signal is consistent with thefurther dispersion (leading to compression in time) upon itssecond traverse of the field line path. The diffuseness ofboth the 1-hop and 2-hop traces is indicative of wave-growth (amplification) and the triggering (and retriggeringduring each equatorial traverse) of emissions, combinedwith multi-path propagation [Helliwell, 1988].[8] Figure 3 shows a 1-min record from the same hour.

The ELF/VLF frequency-time format transmitted is evidentin the top panel, showing Chistochina data from an earliercampaign when the ELF/VLF signals were exceptionallystrong and well defined, as is also evidenced by thepresence of harmonics resulting from the non-linear natureof the ELF/VLF generation in the ionosphere. Evident in themiddle panel of Figure 3 is the 2-hop signal similar to theone seen in the top panel of Figure 2, as well as additionalhops (4th, 6th, 8th, and 10th).[9] Although the Tangaroa data are noisy due to the hum

on the ship, a few of the 1-hop signals showed evidence ofthe triggering of ‘hook’ emissions, as shown in Figure 4.These emissions repeat in multiple minutes in time associ-ation with the 1-hop ramp and exhibit the hook-likefrequency-time signature that is one of the known formsof triggered VLF emissions [Helliwell, 1965, p. 209].When initiated by Siple transmitter signals in this region(4 < L < 5), such emissions are typically preceded bytemporal growth of 10–30 dB [Helliwell, 1988]. Theobservation of triggered emissions is thus strong evidenceof amplification of the injected signals in high altitudeinteraction regions.[10] The absolute magnitudes of the HAARP 1-hop

signal and multi-hop echoes were �0.01 pT to 0.1 pT,substantially smaller than the typical values of HAARPsignals of �1 to �10 pT observed at Chistochina (Moore etal., submitted manuscript, 2004). While the locally observed‘parent’ signals were even weaker, comparing the ampli-tudes of the locally observed signals with the absoluteamplitudes of the 2-hop and higher order echoes does notnecessarily allow a determination of the magnetosphericamplification of the injected ELF/VLF signals. The altitudedistribution of the heated electrojet currents that radiate theELF/VLF signal is not known and is dependent on thealtitude profiles of electron density and electrical conduc-tivity, and the magnitude and altitude distribution of theelectric field.

[11] During the �1 hour period of observation of themulti-hop whistler-mode signals, the HF carrier frequencywas switched from 3.25 MHz to 5.8 MHZ with littlediscernable difference in the magnetospheric response. Nosignificant differences were found in the properties ofHAARP ELF/VLF signals received on the high altitudeCLUSTER satellites [Platino et al., 2004] for HF carrierfrequencies of 3.2 MHz versus 5.8 MHz. The whistler-mode echoes were observed during daylight hours, 1900–2000 HAARP local time, more than an hour before sunset,which on April 20th occurred at approximately 2124 localtime.[12] Whistler-mode signal amplification may often be

limited to an ‘active’ (and sometimes narrow) frequencyrange that is dependent upon the distribution of interactingelectrons and may also be located close to a band of naturalwave activity [Sonwalkar et al., 1997; Carpenter et al.,1997]. Amplification may also be dependent upon thefrequency-time slope of the injected signal, with the moregradual slopes inducing a greater response [Carlson et al.,1985]. Note from Figure 3 that the only component of thetransmitted modulation pattern that leads to whistler-modeechoes is a portion of the 0.5 kHz/s frequency-time rampbetween �1.2 and �2 kHz. It is likely that the 2-s longconstant-frequency pulses did not lead to echoes becausethe pulse at 1225 Hz was just below the active range whilethe pulse at 1875 Hz was close to its upper boundary. In fact,a few rather weak 1-hop signal components of the 1225 Hzpulse were observed near 0330–0331 UT.[13] Geomagnetic conditions during the observations

were generally quiet, with maximum Kp being 2� duringthe past 24 hours, although conditions were disturbed(maximum Kp of 3+ and 4�) on April 16th and less so(maximum Kp of 3�) on the 17th through the 18th.Calibrated auroral electrojet (AE) indices are not yet avail-able but preliminary results show calm conditions, consis-tent with magnetometer readings from the HAARP siteshowing deviations from baseline of <30 nT.

5. Interpretation

[14] Whistler-dispersion analysis was used to determinethe L-shell of propagation and the equatorial cold plasmadensity. The time delay (at each frequency over the range of�1 to �2 kHz) between the time of origin of the originalfrequency-time ramp signal and the leading edge of the echowas measured. The measured data points were extrapolatedto determine the ‘nose’ frequency fn of minimum time delayand hence fHeq, the equatorial gyrofrequency along the fieldline, through the relation fn ’ 0.4fHeq [Sazhin et al., 1992].The measured values were then used together with adiffusive equilibrium model of the cold plasma distributionalong the field line [Angerami and Thomas, 1964] to inferthe equatorial electron density Neq. This analysis revealedvalues of L ’ 4.9 and Neq ’ 280 cm�3, consistent with theempirical model of Carpenter and Anderson [1992].[15] The frequency-time traces of the 1-hop and higher-

order echoes determined by integrating the group velocityalong the field line are superimposed as white traces on thespectrograms in Figure 5a, as a consistency check on thedetermination of L and Neq. The calculated traces agree wellwith the leading edges of the whistler-mode hops, but the

Figure 4. ‘Hook’ emissions triggered by 1-hop rampsignals.

L24805 INAN ET AL.: EXCITATION OF DUCTED ELF/VLF SIGNALS BY HAARP L24805

3 of 4

diffuse extensions that follow the leading edges are clearlyevident.[16] With the L-shell of propagation and the cold plasma

density determined, the energy of the electrons that wouldundergo first order cyclotron resonance interactions with theinjected waves can be calculated [e.g., see Chang and Inan,1983]. Resonant electron energy as a function of geomag-netic latitude along the field line for three different pitchangles is shown in Figure 5b. Since high-pitch-angleelectrons likely drive the gyroresonance instability [Bell etal., 2000], the electrons involved in the amplification of theinjected waves and the triggering of emissions must havehad energies of a few tens of keV.

6. Summary

[17] Observations of the excitation of ducted whistler-mode echoes by modulated HAARP HF transmissionsshow that controlled ELF/VLF wave-injection experimentsaimed at investigating the coherent cyclotron resonancegrowth, amplification and emission triggering processes inthe magnetosphere can be conducted with this facility.In view of the demonstrated (Moore et al., submittedmanuscript, 2004) capabilities of HAARP in exciting wavesover a very broad range of frequencies (from a few Hz to30 kHz), and since the HAARP facility is currently beingupgraded in radiated power by a factor of >3, such experi-ments can provide an excellent test bed for study ofmagnetospheric wave-particle interactions.

[18] Acknowledgments. This work was supported by the High-frequency Active Auroral Research Program (HAARP), the DefenseAdvanced Research Programs Agency (DARPA), and by the Office ofNaval Research (ONR) via ONR grant N00014-03-0631 to StanfordUniversity. Special thanks are due to the National Institute for Water andAtmospheric research (NIWA) of New Zealand and the crew of RVTangaroa for their outstanding support during ship operations, to our hostsDoyle and Norma Traw at the Chistochina B&B, and Morris Cohen andJustin Tan for their work on the ELF/VLF receiver used on RV Tangaroa.

ReferencesAngerami, J. J., and J. O. Thomas (1964), Studies of planetary atmospheres:1. The distribution of ions and electrons in the Earth’s ionosphere,J. Geophys. Res., 69, 4537.

Barr, R., and P. Stubbe (1984), ELF and VLF radiation from the polarelectrojet antenna, Radio Sci., 19, 1111.

Barr, R., and P. Stubbe (1991), ELF radiation from the Tromsø ‘‘superheater’’ facility, Geophys. Res. Lett., 18, 1035.

Bell, T. F., U. S. Inan, R. A. Helliwell, and J. D. Scudder (2000), Simulta-neous triggered VLF emissions and energetic electron distributions ob-served on POLAR with PWI and HYDRA, Geophys. Res. Lett., 27, 165.

Carlson, C. R., R. L. Helliwell, and D. L. Carpenter (1985), Variablefrequency VLF signals in the magnetosphere: Associated phenomenaand plasma diagnostics, J. Geophys. Res., 90, 1507.

Carpenter, D. L., and R. R. Anderson (1992), An ISEE/whistler model ofequatorial electron density in the magnetosphere, J. Geophys. Res., 97,1097.

Carpenter, D. L., and D. M. Sulic (1988), Ducted whistler propagationoutside the plasmapause, J. Geophys. Res., 93, 9731.

Carpenter, D. L., V. S. Sonwalkar, R. A. Helliwell, M. Walt, U. S. Inan,D. L. Caudle, and M. Ikeda (1997), Probing properties of the magneto-spheric hot plasma distribution by whistler mode wave injection at multi-ple frequencies: Evidence of spatial as well as temporal wave growth,J. Geophys. Res., 102, 14,355.

Chang, H. C., and U. S. Inan (1983), Quasi-relativistic electron precipita-tion due to interactions with coherent VLF waves in the magnetosphere,J. Geophys. Res., 88, 282.

Ferraro, A., H. Lee, R. Allshouse, K. Carroll, A. Tomko, F. Kelly, andR. Joiner (1982), VLF/ELF radiation from dynamo current systemmodulated by powerful HF signals, J. Atmos. Terr. Phys., 44, 1113.

Helliwell, R. A. (1965), Whistlers and Related Ionospheric Phenomena,Stanford Univ. Press, Stanford, Calif.

Helliwell, R. A. (1988), VLF wave simulation experiments in the magneto-sphere from Siple Station, Antarctica, Rev. Geophys., 26, 551.

Milikh, G. M., K. Papadopoulos, M. McMarrick, and J. Preston (1999),ELF emission generated by the HAARP HF-heater using varying fre-quency and polarization, Izv. Vyssh. Uchebn. Zaved Radiofiz., 42(8), 728.

Platino, M., U. S. Inan, T. F. Bell, J. Pickett, E. J. Kennedy, J. G. Trotignon,J. L. Rauch, and P. Canu (2004), Cluster observations of ELF/VLF sig-nals generated by modulated heating of the lower ionosphere with theHAARP HF transmitter, Ann. Geophys., 22, 2643.

Rietveld, M. T., P. Stubbe, and H. Kopka (1989), On the frequency depen-dence of ELF/VLF waves produced by modulated ionospheric heating,Radio Sci., 24, 270.

Sazhin, S. S., M. Hayakawa, and K. Bullough (1992), Whistler diagnosticsof magnetospheric parameters: A review, Ann. Geophys., 10, 293.

Sonwalkar, V. S., D. L. Carpenter, R. A. Helliwell, M. Walt, U. S. Inan,D. L. Caudle, and M. Ikeda (1997), Properties of the magnetospherichot plasma distribution deduced from whistler mode wave injection at2400 Hz: Ground based detection of azimuthal structure in magneto-spheric hot plasmas, J. Geophys. Res., 102, 14,363.

Starks, M. J., M. C. Lee, and P. Jastrzebski (2001), Interhemispheric pro-pagation of VLF transmissions in the presence of ionospheric HF heating,J. Geophys. Res., 106, 5579.

Stubbe, P., H. Kopka, M. T. Rietveld, and R. L. Dowden (1982), ELF andVLF wave generation by modulated HF heating of the current carryinglower ionosphere, J. Atmos. Terr. Phys., 44, 1123.

Villasenor, J., A. Wong, B. Song, J. Pau, M. McCarrick, and D. Sentman(1996), Comparison of ELF/VLF generation modes in the ionosphere bythe HIPAS heater array, Radio Sci., 31, 211.

�����������������������T. F. Bell, D. L. Carpenter, M. Gol-kowski, R. C. Moore, and U. S. Inan,

STAR Laboratory, Stanford University, Packard Bldg., Rm. 355, 350 SerraMall, Stanford, CA 94305, USA. (inan@nova.stanford.edu)E. Kennedy, Naval Research Laboratory, Code 5550, 4555 Overlook

Ave., SW, Washington, DC 20375, USA.P. Kossey, Air Force Research Laboratory, AFRL/USBI, 29 Randolph

Rd., Hanscom AFB, MA 01731-3010, USA.S. Z. Meth, DARPA Tactical Technology Office, 3701 Fairfax Dr.,

Arlington, VA 22203-1714, USA.E. Paschal, Whistler Radio Services, 8910 Villa Beach Rd., Anderson

Island, WA 98303, USA.N. Reddell, 87 Ludlow St., Saratoga Springs, NY 12866, USA.

Figure 5. (a) Overlay on observed multi-hop signals tracesshowing expected dispersion that would be experienced byHAARP transmitted signals. (b) First order resonantelectron energies for the observed data for various pitchangles.

L24805 INAN ET AL.: EXCITATION OF DUCTED ELF/VLF SIGNALS BY HAARP L24805

4 of 4